PROJECT SUMMARY/ABSTRACT Dialysis-requiring acute kidney injury (AKI-D) is one of the most serious complications for hospitalized patients. Renal recovery after AKI-D ? defined as return of sufficient native kidney function to come off dialysis ? is a critically important clinical and patient-oriented outcome. However, we currently have limited ability to predict recovery in an individual AKI-D patient, and there are no interventions proven to improve the likelihood of recovery. Our overall goal is to improve outcomes in patients with AKI-D. We propose to develop and validate a clinical prediction model for recovery after AKI-D using an established database from a large integrated health-care delivery system. Guided by our prediction model, we will also conduct a small, single-center pilot randomized controlled trial among AKI-D patients with a reasonable chance of recovery to test the feasibility and safety of a conservative dialysis strategy, in which dialysis is not continued unless specific metabolic or clinical indications for dialysis are present. We hypothesize that, compared to the conventional practice of thrice-weekly dialysis, a conservative dialysis strategy will reduce the number of acute hemodialysis sessions needed and improve the likelihood of renal recovery. This pilot study will play several key roles in the design of a larger future trial by identifying potential barriers to recruitment and affording the opportunity to optimize intervention delivery. Our specific aims are: Aim 1. To derive a clinical prediction model to predict the chances of renal recovery after AKI-D by 90 days after initiation of renal replacement therapy. Aim 2. Among AKI-D patients classified as having >30% chance of recovery, to determine the feasibility/safety of and adherence to a strategy of conservative dialysis strategy in a single-center pilot randomized clinical trial (n=20).